More Like this

1. Where It Really Matters: Few-Shot Environmental Conservation Media Monitoring for Low-Resource Languages

   https://doi.org/10.1609/aaai.v38i20.30218  

   Jain, Sameer; Keh, Sedrick Scott; Chhetri, Shova; Dewan, Karun; Izquierdo, Pablo; Prussmann, Johanna; Shrestha, Pooja; Suárez, César; Shi, Zheyuan Ryan; Li, Lei; et al (March 2024, Proceedings of the AAAI Conference on Artificial Intelligence)

   Environmental conservation organizations routinely monitor news content on conservation in protected areas to maintain situational awareness of developments that can have an environmental impact. Existing automated media monitoring systems require large amounts of data labeled by domain experts, which is only feasible at scale for high-resource languages like English. However, such tools are most needed in the global south where the news of interest is mainly in local low-resource languages, and far fewer experts are available to annotate datasets on a sustainable basis. In this paper, we propose NewsSerow, a method to automatically recognize environmental conservation content in low-resource languages. NewsSerow is a pipeline of summarization, in-context few-shot classification, and self-reflection using large language models (LLMs). Using at most 10 demonstration example news articles in Nepali, NewsSerow significantly outperforms other few-shot methods and can achieve comparable performance with models fully fine-tuned using thousands of examples. With NewsSerow, Organization X has been able to deploy the media monitoring tool in Nepal, significantly reducing their operational burden, and ensuring that AI tools for conservation actually reach the communities that need them the most. NewsSerow has also been deployed for countries with other languages like Colombia. 

   more » « less
2. Areal Analysis of Language Attitudes and Practices: A Case Study from Nepal

   Hildebrandt, Kristine A.; Hu, Shunfu (January 2017, Language documentation and conservation)

   This paper has two aims. One aim is to consider non-structural (language attitude and use) variables as valid in the field of dialect and linguistic geography in an inner Himalayan valley of Nepal, where four languages have traditionally co- existed asymmetrically and which demonstrate different degrees of vitality vs. endangerment. The other aim is an application of modified spatiality as it aligns with speaker attitudes and practices amidst recent and ongoing socio-economic and population changes. We demonstrate that variation in self-reported attitudes and practices across languages in this region can be explained as much with adjusted spatial factors (labeled ‘social space’) as with traditional social factors (e.g. gender, age, formal education, occupation, etc.). As such, our study contributes to a dis- course on the role and potential of spatiality in sociolinguistic analyses of smaller language communities. 

   more » « less
3. A Sociolinguistic Survey of the Languages of Manang, Nepal: Co-Existence and Endangerment

   Hildebrandt, Kristine A; Dhakal, Dubi Nanda; Bond, Oliver; Vallejo, Matthew; Fyffe, Andrea (January 2015, Journal of Indigenous Nationalities)

   With over one hundred languages from four major families (and at least one isolate), and a similarly high number of caste-clan and ethnic groupings, Nepal is a country of undisputed ethno-linguistic diversity (CBS 2012; Kansakar 2006; Gurung 1998). It is also a country of increasingly rapid social, cultural, political and economic change with ensuing geographic movement and language displacement (Angdembe 2013; Rai 2013; Tumbahang 2012). The purpose of this study is to examine the intersection of traditional language practices and cultural diversity with these modern changes as they are relevant for a group of four languages spoken inside of the political boundaries of Manang, Nepal. 

   more » « less
4. Cross-lingual articulatory feature information transfer for speech recognition using recurrent progressive neural networks

   https://doi.org/10.21437/Interspeech.2022-11202  

   Morshed, Mahir; Hasegawa-Johnson, Mark (September 2022, Proceedings of Interspeech 2022)

   A system for the lateral transfer of information from end-to-end neural networks recognizing articulatory feature classes to similarly structured networks recognizing phone tokens is here proposed. The system connects recurrent layers of feature detectors pre-trained on a base language to recurrent layers of a phone recognizer for a different target language, this inspired primarily by the progressive neural network scheme. Initial experiments used detectors trained on Bengali speech for four articulatory feature classes—consonant place, consonant manner, vowel height, and vowel backness—attached to phone recognizers for four other Asian languages (Javanese, Nepali, Sinhalese, and Sundanese). While these do not currently suggest consistent performance improvements across different low-resource settings for target languages, irrespective of their genealogic or phonological relatedness to Bengali, they do suggest the need for further trials with different language sets, altered data sources and data configurations, and slightly altered network setups. 

   more » « less
5. Larger spleens and greater splenic contraction during exercise may be an adaptive characteristic of Nepali Sherpa at high‐altitude

   https://doi.org/10.1002/ajhb.24090  

   Brutsaert, Tom D.; Harman, Taylor Shay; Bigham, Abigail W.; Kalker, Anne; Jorgensen, Kelsey C.; Zhu, Kimberly T.; Steiner, Bethany C.; Hawkins, Ella; Day, Trevor A.; Kunwar, Ajaya J.; et al (May 2024, American Journal of Human Biology)

   Abstract ObjectivesThe Sherpa ethnic group living at altitude in Nepal may have experienced natural selection in response to chronic hypoxia. We have previously shown that Sherpa in Kathmandu (1400 m) possess larger spleens and a greater apnea‐induced splenic contraction compared to lowland Nepalis. This may be significant for exercise capacity at altitude as the human spleen responds to stress‐induced catecholamine secretion by an immediate contraction, which results in transiently elevated hemoglobin concentration ([Hb]). MethodsTo investigate splenic contraction in response to exercise at high‐altitude (4300 m; P_b = ~450 Torr), we recruited 63 acclimatized Sherpa (29F) and 14 Nepali non‐Sherpa (7F). Spleen volume was measured before and after maximal exercise on a cycle ergometer by ultrasonography, along with [Hb] and oxygen saturation (SpO₂). ResultsResting spleen volume was larger in the Sherpa compared with Nepali non‐Sherpa (237 ± 62 vs. 165 ± 34 mL,p < .001), as was the exercise‐induced splenic contraction (Δspleen volume, 91 ± 40 vs. 38 ± 32 mL,p < .001). From rest to exercise, [Hb] increased (1.2 to 1.4 g.dl⁻¹), SpO₂decreased (~9%) and calculated arterial oxygen content (CaO₂) remained stable, but there were no significant differences between groups. In Sherpa, both resting spleen volume and the Δspleen volume were modest positive predictors of the change (Δ) in [Hb] and CaO₂with exercise (p‐values from .026 to .037 and R²values from 0.059 to 0.067 for the predictor variable). ConclusionsLarger spleens and greater splenic contraction may be an adaptive characteristic of Nepali Sherpa to increase CaO₂during exercise at altitude, but the direct link between spleen size/function and hypoxia tolerance remains unclear. 

   more » « less